Search results for "Convolution"

showing 10 items of 334 documents

PNeuro: A scalable energy-efficient programmable hardware accelerator for neural networks

2018

Proceedings of a meeting held 19-23 March 2018, Dresden, Germany; International audience; Artificial intelligence and especially Machine Learning recently gained a lot of interest from the industry. Indeed, new generation of neural networks built with a large number of successive computing layers enables a large amount of new applications and services implemented from smart sensors to data centers. These Deep Neural Networks (DNN) can interpret signals to recognize objects or situations to drive decision processes. However, their integration into embedded systems remains challenging due to their high computing needs. This paper presents PNeuro, a scalable energy-efficient hardware accelerat…

Neural network hardwareComputer sciencePooling02 engineering and technologyLow power0202 electrical engineering electronic engineering information engineeringSIMDField-programmable gate arrayFPGAComputer architecturesRoutingArtificial neural networkASIC[SCCO.NEUR]Cognitive science/Neuroscience020208 electrical & electronic engineering[SCCO.NEUR] Cognitive science/NeuroscienceField programmable gate arraysConvolution020202 computer hardware & architectureGeneratorsComputer architectureScalabilityHardware accelerationRouting (electronic design automation)Neural networksEfficient energy use
researchProduct

Some experimental issues of AFM tip blind estimation. The effect of noise and resolution

2006

The convolution of tip shape on sample topography can introduce significant inaccuracy in an AFM image, when the tip radius is comparable to the typical dimension of the sample features to be observed. The blind estimation method allows one to obtain information on the AFM tip through an unknown characterizer sample and thus to perform the deconvolution of the tip shape from an image. When applying the blind estimation method to determine the AFM tip shape, some apparently trivial issues relating to the experimental operating parameters must be taken into account. In this paper, the effects of the operating parameters, e.g., sampling intervals (resolution) and instrumental noise, have been …

Noise (signal processing)Applied MathematicsAcousticsResolution (electron density)Sampling (statistics)atomic force microscopy tip characterization blind estimationRadiusSample (graphics)ConvolutionDimension (vector space)StatisticsDeconvolutionInstrumentationEngineering (miscellaneous)Mathematics
researchProduct

Infinitely Divisible Distributions

2020

For every n, the normal distribution with expectation μ and variance σ 2 is the nth convolution power of a probability measure (namely of the normal distribution with expectation μ/n and variance σ 2/n). This property is called infinite divisibility and is shared by other probability distributions such as the Poisson distribution and the Gamma distribution. In the first section, we study which probability measures on the real line are infinitely divisible and give an exhaustive description of this class of distributions by means of the Levy–Khinchin formula.

Normal distributionCombinatoricssymbols.namesakesymbolsGamma distributionProbability distributionPoisson distributionConvolution powerInfinite divisibilityStable distributionProbability measureMathematics
researchProduct

A Comparative Analysis of Residual Block Alternatives for End-to-End Audio Classification

2020

Residual learning is known for being a learning framework that facilitates the training of very deep neural networks. Residual blocks or units are made up of a set of stacked layers, where the inputs are added back to their outputs with the aim of creating identity mappings. In practice, such identity mappings are accomplished by means of the so-called skip or shortcut connections. However, multiple implementation alternatives arise with respect to where such skip connections are applied within the set of stacked layers making up a residual block. While residual networks for image classification using convolutional neural networks (CNNs) have been widely discussed in the literature, their a…

Normalization (statistics)General Computer ScienceComputer scienceFeature extractionESC02 engineering and technologycomputer.software_genreResidualConvolutional neural networkconvolutional neural networks0202 electrical engineering electronic engineering information engineeringGeneral Materials Scienceurbansound8kAudio signal processingBlock (data storage)Contextual image classificationGeneral EngineeringAudio classification020206 networking & telecommunications113 Computer and information sciences020201 artificial intelligence & image processinglcsh:Electrical engineering. Electronics. Nuclear engineeringData mininglcsh:TK1-9971computerresidual learningIEEE Access
researchProduct

Boosting background suppression in the NEXT experiment through Richardson-Lucy deconvolution

2021

The NEXT collaboration: et al.

Nuclear and High Energy PhysicsIonizationPhysics - Instrumentation and DetectorsIonitzacióFOS: Physical sciencesdouble beta decayRichardson–Lucy deconvolutionBragg peakElectronQC770-79801 natural sciencesSignalHigh Energy Physics - ExperimentHigh Energy Physics - Experiment (hep-ex)IonizationDouble beta decayNuclear and particle physics. Atomic energy. Radioactivitygas0103 physical sciences010306 general physicsPhysics010308 nuclear & particles physicsRaigs beta -- DesintegracióInstrumentation and Detectors (physics.ins-det)Computational physicsdark matter and double beta decay (experiments)Beta rays -- DecayDeconvolutionEnergy (signal processing)
researchProduct

Demonstration of background rejection using deep convolutional neural networks in the NEXT experiment

2021

[EN] Convolutional neural networks (CNNs) are widely used state-of-the-art computer vision tools that are becoming increasingly popular in high-energy physics. In this paper, we attempt to understand the potential of CNNs for event classification in the NEXT experiment, which will search for neutrinoless double-beta decay in Xe-136. To do so, we demonstrate the usage of CNNs for the identification of electron-positron pair production events, which exhibit a topology similar to that of a neutrinoless double-beta decay event. These events were produced in the NEXT-White high-pressure xenon TPC using 2.6 MeV gamma rays from a Th-228 calibration source. We train a network on Monte Carlo-simulat…

Nuclear and High Energy PhysicsPhysics - Instrumentation and DetectorsCalibration (statistics)Computer Science::Neural and Evolutionary ComputationNuclear physicsFOS: Physical sciencesTopology (electrical circuits)01 natural sciencesConvolutional neural networkAtomicPartícules (Física nuclear)High Energy Physics - ExperimentInteraccions electró-positróTECNOLOGIA ELECTRONICAHigh Energy Physics - Experiment (hep-ex)Particle and Plasma PhysicsDouble beta decay0103 physical sciencesDark Matter and Double Beta Decay (experiments)NuclearNuclear Matrixlcsh:Nuclear and particle physics. Atomic energy. Radioactivity010306 general physicsElectron-positron interactionsMathematical PhysicsParticles (Nuclear physics)PhysicsQuantum Physics010308 nuclear & particles physicsbusiness.industryEvent (computing)Network onSIGNAL (programming language)MolecularFísicaPattern recognitionDetectorInstrumentation and Detectors (physics.ins-det)Beta DecayDouble beta decayNuclear & Particles PhysicsDoble desintegració betaIdentification (information)lcsh:QC770-798Física nuclearArtificial intelligencebusinessJournal of High Energy Physics
researchProduct

Sensitivity enhancement in pulse EPR distance measurements

2004

Established pulse EPR approaches to the measurement of small dipole-dipole couplings between electron spins rely on constant-time echo experiments to separate relaxational contributions from dipolar time evolution. This requires a compromise between sensitivity and resolution to be made prior to the measurement, so that optimum data are only obtained if the magnitude of the dipole-dipole coupling is known beforehand to a good approximation. Moreover, the whole dipolar evolution function is measured with relatively low sensitivity. These problems are overcome by a variable-time experiment that achieves suppression of the relaxation contribution by reference deconvolution. Theoretical and exp…

Nuclear and High Energy PhysicsProtein ConformationBiophysicsAnalytical chemistryBiochemistrySensitivity and Specificitylaw.inventionlawspin labelingSensitivity (control systems)protein structurepair correlation functionElectron paramagnetic resonanceCouplingSpinsChemistryPulsed EPRRelaxation (NMR)Time evolutionElectron Spin Resonance SpectroscopyPhotosystem II Protein ComplexReproducibility of ResultsSignal Processing Computer-AssistedELDORCondensed Matter PhysicsComputational physicsDeconvolutionEPRAlgorithms
researchProduct

On the condition number of the antireflective transform

2010

Abstract Deconvolution problems with a finite observation window require appropriate models of the unknown signal in order to guarantee uniqueness of the solution. For this purpose it has recently been suggested to impose some kind of antireflectivity of the signal. With this constraint, the deconvolution problem can be solved with an appropriate modification of the fast sine transform, provided that the convolution kernel is symmetric. The corresponding transformation is called the antireflective transform. In this work we determine the condition number of the antireflective transform to first order, and use this to show that the so-called reblurring variant of Tikhonov regularization for …

Numerical AnalysisAlgebra and Number TheoryBoundary conditionsTikhonov regularizationMathematical analysisDeconvolutionUpper and lower boundsRegularization (mathematics)ConvolutionTikhonov regularizationTransformation (function)Discrete Mathematics and CombinatoricsApplied mathematicsFast sine transformGeometry and TopologyUniquenessDeconvolutionCondition numberAntireflective transformMathematicsLinear Algebra and its Applications
researchProduct

Implicit-explicit methods for a class of nonlinear nonlocal gradient flow equations modelling collective behaviour

2019

Abstract The numerical solution of nonlinear convection-diffusion equations with nonlocal flux by explicit finite difference methods is costly due to the local spatial convolution within the convective numerical flux and the disadvantageous Courant-Friedrichs-Lewy (CFL) condition caused by the diffusion term. More efficient numerical methods are obtained by applying second-order implicit-explicit (IMEX) Runge-Kutta time discretizations to an available explicit scheme for such models in Carrillo et al. (2015) [13] . The resulting IMEX-RK methods require solving nonlinear algebraic systems in every time step. It is proven, for a general number of space dimensions, that this method is well def…

Numerical AnalysisApplied MathematicsNumerical analysisCPU timeSpace (mathematics)Computer Science::Numerical AnalysisMathematics::Numerical AnalysisConvolutionTerm (time)Computational MathematicsNonlinear systemApplied mathematicsBalanced flowReduction (mathematics)MathematicsApplied Numerical Mathematics
researchProduct

Differentiating Malignant from Benign Pigmented or Non-Pigmented Skin Tumours—A Pilot Study on 3D Hyperspectral Imaging of Complex Skin Surfaces and …

2022

Several optical imaging techniques have been developed to ease the burden of skin cancer disease on our health care system. Hyperspectral images can be used to identify biological tissues by their diffuse reflected spectra. In this second part of a three-phase pilot study, we used a novel hand-held SICSURFIS Spectral Imager with an adaptable field of view and target-wise selectable wavelength channels to provide detailed spectral and spatial data for lesions on complex surfaces. The hyperspectral images (33 wavelengths, 477–891 nm) provided photometric data through individually controlled illumination modules, enabling convolutional networks to utilise spectral, spatial, and skin-surface mo…

OPTICAL COHERENCE TOMOGRAPHYskin cancerhyperspectral imagingskin imagingphotometric stereoMELANOMAGeneral Medicineneuroverkotdiagnostiikkabiomedical optical imagingnon-invasive imagingDIAGNOSISCANCERoptical modellingkarsinoomatCLASSIFICATIONihosyöpäkoneoppiminenSDG 3 - Good Health and Well-beingbiomedical optical imaging; convolutional neural networks; hyperspectral imaging; non-invasive imaging; optical modelling; photometric stereo; skin cancer; skin imaging3121 General medicine internal medicine and other clinical medicineconvolutional neural networks/dk/atira/pure/sustainabledevelopmentgoals/good_health_and_well_beingmelanoomahyperspektrikuvantaminen
researchProduct